5_5_RAG.py

from langchain_google_genai import ChatGoogleGenerativeAI
from langchain.vectorstores import Chroma
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.chains import RetrievalQA
from dotenv import load_dotenv
import os

load_dotenv()

chat = ChatGoogleGenerativeAI(
    model="gemini-1.5-flash",  # Or gemini-1.5-pro etc.
    temperature=0.7,
    google_api_key=os.getenv("GEMINI_API_KEY"),  # pass key directly or rely on env
)

embeddings = HuggingFaceEmbeddings(
    model_name="all-MiniLM-L6-v2",
)

db = Chroma(
    persist_directory="embeddings_db",  # Directory to store the embeddings
    embedding_function=embeddings
)

retriever = db.as_retriever()

chain = RetrievalQA.from_chain_type(
    llm=chat,
    retriever=retriever,
    chain_type="stuff"
)

question = "What do you know about Issac Newton?"

result = chain.run(question)

print("Result:", result)